Analytical and clinical evaluations of SNIBE Maglumi chemiluminescent immunoassay for the detection of SARS-CoV-2 antigen in salivary samples

Objectives: In this study, we describe the analytical and clinical performances of the SNIBE Maglumi SARS-CoV-2 antigen fully-automated chemiluminescent immunoassay (MAG-CLIA) on salivary samples. Methods: Limit of detection (LOD), linearity and precision were tested for values close to or below the declared LOD. Clinical performance of MAG-CLIA was evaluated on left- over salivary samples from the healthcare workers (HCW) surveillance program, at the University-Hospital of Padova. Salivary samples were analyzed by Lumipulse G SARS-CoV-2 Ag, and in case where the values exceeded 0.41 ng/L, further testing was conducted using TaqPathTM COVID-19 RT-PCR (Applied Biosystems, Thermo Fisher Scientific). Results: The estimated MAG-CLIA LOD was 3ng/L, with repeatability of 7.5 %. Good linearity was demonstrated by diluting two samples at 52.7 ng/L and 211.4 ng/L. Of the 228 HCW samples, 59/228 (25.9 %) were positive, 169/228 (74.1 %) were negative. MAG-CLIA SARS-CoV-2 sAg median level (and interquartile range [IQR]) was 5.03 ng/L (<0.001–35.8 ng/L) for positive and <0.001 ng/L (<0.001 ng/L) for negative sam- ples. MAG-CLIA AUC was 0.795 (95 % CI: 0.720–0.871). Using the best cut-off, 3.5ng/L, sensitivity and specificity were 57.1 % (95 % CI: 42.2–71.2 %) and 97.0 % (95 % CI: 93.2–99.0 %), respectively. The agreement with the molecular assay was 88.1 % (Cohen’s kappa 0.606 [SE=0.066, p<0.001]). Conclusions: TheanalyticalperformancesofMAG-CLIAare satisfactory, also when values below LOD were tested. In saliva samples, although specificity was elevated, clinical performance was not comparable with that on nasopha- ryngeal swabs (NPS).